Eden AI

@edenai what's my remaining credit balance and how much did I spend on OpenAI calls this month?

Your Eden AI account has 487.32 credits remaining. From January 1–31, you consumed 112.68 credits on OpenAI models: GPT-4o accounted for 89.40 credits (text generation), GPT-4o-mini for 14.12 credits (chat completions), and Whisper for 9.16 credits (audio transcription). No other providers were billed in this period. Next renewal is February 1.

Shows the cost-overview tool filtering by provider and date range. Helpful for budget tracking when you're routing requests through multiple AI vendors via Eden AI. The MCP requires an API key with billing-read scope. Credit balances update in near-real-time, but provider breakdowns may lag by a few minutes during high-volume usage.

When Eden AI wins for embedding vendor evaluation

A 6-person data team building semantic search needs to benchmark three embedding models (OpenAI, Cohere, Voyage) on their internal docs before committing to one vendor. Eden AI's Create Embeddings tool lets them run all three from a single API key and compare vectors side-by-side in the same Switchy thread. The Get Cost Overview tool tracks spend per provider in real time, so they know exactly what each experiment costs before scaling. This works well for evaluation sprints under 500k tokens; beyond that, you'll want direct vendor contracts for volume pricing. If your team already knows which embedding model to use, skip Eden AI and go straight to that provider's MCP—you'll save the markup and the abstraction layer. But for the 'which vendor should we pick' phase, Eden AI collapses three integrations into one and gives you apples-to-apples cost data in a single workspace.

When Eden AI helps track multi-provider support spend

A 3-person support team uses GPT-4 for complex tickets and Claude for routine triage, but they have no visibility into which model is burning budget. Eden AI's Create Chat Completion and Create Anthropic Message tools return per-call cost metadata, and Get Cost Overview aggregates it by provider and date range. In Switchy, the team runs both models from the same thread and sees cumulative spend in a single dashboard—no need to log into OpenAI and Anthropic separately. This setup shines when you're running 2-4 models in parallel and need unified billing insight. If you're only using one LLM, Eden AI adds overhead you don't need; just use that vendor's native MCP. And if you're processing thousands of tickets daily, Eden AI's per-call markup will exceed the value of consolidated reporting. For small teams in the 'figure out our LLM mix' phase, though, Eden AI turns cost tracking from a spreadsheet chore into a built-in feature.

When Count Anthropic Tokens prevents runaway costs

A 4-person product team uses Claude to summarize user interviews, but they keep hitting rate limits because they don't know how many tokens each interview transcript consumes before sending it. Eden AI's Count Anthropic Tokens tool lets them check token counts in Switchy before calling Create Anthropic Message, so they can batch interviews into chunks that stay under the model's context window. This prevents wasted API calls and surprise overage charges. The tool works best when you're prototyping workflows with variable-length inputs—interview transcripts, support tickets, meeting notes—and need to stay within a monthly token budget. If your inputs are predictable (like 500-word product descriptions), you don't need pre-flight token counting; just estimate once and move on. And if you're already using Anthropic's native MCP, it likely has token counting built in. But for teams juggling multiple LLM providers through Eden AI and trying to forecast costs before execution, this tool turns guesswork into data.